Running a seed node · Radicle docs

To improve data availability of Radicle's peer-to-peer network, participants in the network can choose to act as seeds. Seeds are always-on nodes that automatically track discovered projects, thereby increasing the availability of these projects on the network. Conceptually, they are similar to pubs in Secure Scuttlebutt.

Running a seed node is for contributing to the availability and redundancy of Radicle's peer-to-peer network. Running a seed node is not related to token incentives at this time. For more on how the Radicle token works, please refer to the Introducing RAD post.

To build and run a seed node, you'll have to install some prerequisites on your machine first:

Rust toolchain, you'll need Rust nightly, easiest way to set it up is via rustup
yarn

Make sure to restart terminal after installation

Next, clone and set up the radicle-bins repository:

git clone https://github.com/radicle-dev/radicle-bins.git
cd radicle-bins

Install UI depependencies and build the UI:

(cd seed/ui && yarn && yarn build)

Next, let's set up a directory where the seed can store its data. This is important in case you are running a Radicle Upstream client on the same machine. The default path for both Radicle Upstream and the seed node would otherwise be the same and could result in unexpected behaviour.

mkdir -p ~/.radicle-seed

Then, you'll have to generate a private key:

cargo run -p radicle-keyutil -- --filename ~/.radicle-seed/secret.key

In order for people to connect to the seed node from the internet, you'll have to allow incomming connections for the following ports:

UDP:12345 - for peer data exchange
TCP:80 - for the seed node UI

For this example, let's assume that we have the public IP address 1.2.3.4 at our disposal and that our router is set up to forward the ports UDP:12345 and TCP:80 to the machine where the seed node will be running.

While not strictly necessary, it's nice to have a domain name set up for the public IP address as well. For this example, we'll use this record: seed.example.com. A 1.2.3.4.

Now you're ready to start the seed node. Let's configure it to listen on ports 12345 and 80 on all interfaces, the private key is supplied via STDIN:

cargo run -p radicle-seed-node --release -- \
  --root ~/.radicle-seed \
  --peer-listen 0.0.0.0:12345 \
  --http-listen 0.0.0.0:80 \
  --name "seedling" \
  --public-addr "seed.example.com:12345" \
  --assets-path seed/ui/public \
  < ~/.radicle-seed/secret.key

Note: we configured the seed to use ~/.radicle-seed as its data directory with the --root option. Remember to adjust --public-addr and --name to your setup. --name will be shown as a heading and --public-addr will appear in the seed address as <SEED-ID>@<PUBLIC-ADDR>:<PORT>.

This is what you'll see in the terminal when starting the seed node:

$ cargo run -p radicle-seed-node --release -- \
  --root ~/.radicle-seed \
  --peer-listen 0.0.0.0:12345 \
  --http-listen 0.0.0.0:80 \
  --name "seedling" \
  --public-addr "seed.example.com:12345" \
  --assets-path seed/ui/public \
  < ~/.radicle-seed/secret.key
    Finished release [optimized] target(s) in 0.19s
     Running `target/release/radicle-seed-node --root /Users/rudolfs/.radicle-seed --peer-listen '0.0.0.0:12345' --http-listen '0.0.0.0:80' --name seedling --public-addr 'seed.example.com:12345' --assets-path seed/ui/public`
Nov 12 10:48:03.758  INFO radicle_seed: Initializing tracker to track everything..
Nov 12 10:48:03.758  INFO Protocol::run{local.id=hyy5s7ysg96fqa91gbe7h38yddh4mkokft7y4htt8szt9e17sxoe3h local.addr=0.0.0.0:12345}: librad::net::protocol: Listening
Nov 12 10:48:03.760  INFO Server::run{addr=0.0.0.0:80}: warp::server: listening on http://0.0.0.0:80
Nov 12 10:48:03.760  INFO radicle_seed_node::frontend: Listening(0.0.0.0:12345)

Now point your browser to the address you set instead of http://seed.example.com. This is the seed node dashboard.

Seed node UI

For Upstream clients to connect to your new seed, you'll need to share the seed address. This address can be found in the UI under the name of the seed. In our example the address is:

hyy5s7ysg96fqa91gbe7h38yddh4mkokft7y4htt8szt9e17sxoe3h@seed.example.com:12345

Have a look at the Adding a custom seed node section for more information on how to set up the new seed in Upstream.

Requirements for running a seed node

The requirements for the seed server are minimal. You should be able to run it with just 1GB of RAM and a single core server.

The two main resources to watch out for are the storage and network quotas/cost. The storage scales directly with the amount of code in all projects stored. As of mid-2021 you won't need more than 10G of storage, but better plan for quick expansion.

The network should sustain 300kbps and your monthly transfer quotas should be able to cover that.

Common questions

Why is my node is not connecting to others or cloning any projects?

By default your node will not automatically join any network, however it will learn about other accessible nodes from connected clients. To initiate the connection automatically, you can start the node with the --bootstrap option and a list of seed node addresses.

Are there any rewards for running the seed node?

Not at the moment. This is technically hard to implement fairly, but you're welcome to discuss ideas for it on the forum.

Why is the seed host running out of resources?

There are still issues to solve in the code. In order to keep your host responsive and handle temporary problems and traffic spikes, it's best to limit the resources available to it. On Linux this can be done with cgroups. Limiting available memory to below 1GB and maximum CPU usage to 80% will allow you to gracefully restart the service if needed.

Upgrading a seed node from v0.1.x to v0.2.x

Get the latest changes on radicle-bins:

cd radicle-bins
git checkout master
git pull

The data layout between v0.1 and v0.2 seed nodes changed, this means you'll have to remove the old seed data directory and re-seed all of your projects from source. Assuming you followed this documentation to set up your seed node it should be sufficient to remove the directory where --root pointed to.
Set up and run the seed node as per the steps outlined above starting with creating a new key.